“

“Logic will get you from A to B.

Imagination will take you everywhere.”
 Lecture Set 4:
Optical Communication Systems

Thas A Nirmalathas
nirmalat @ unimelb . edu . Au

Contact me with a subject line starting

[UoJGuestLecture]
 Frequency Chirp
• When the laser current is directly modulated, the laser may experience
dynamic line broadening

• Changes in the carrier density in the active region result in changes in the
refractive index, which leads to changes in the instantaneous optical
frequency - frequency chirp

• When the injection current is varied

Þ Carrier density varies
ÞRefractive index varies
ÞEmission wavelength varies with time
ÞFrequency Chirp

• Combined with fibre dispersion, frequency chirp limits the speed and and length of
transmission of optical signals
Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide3
 Red and Blue Frequency Shifts
• Two main areas of interest - turn-on and turn-off, where the carrier density
changes significantly
– turn-on -- N(t) > Nth , so f(t) > fss Þ BLUE SHIFT (to lower wavelengths)
– turn-off -- N(t) < Nth , so f(t) < fss Þ RED SHIFT (to higher wavelengths)

BLUE
SHIFT

RED
SHIFT

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide4
 Linewidth Enhancement Factor

• Measure of refractive index change with modulation current

– rate of phase change over time, relative to normalised rate of optical power
change over time
df
a lw = dt
dP Linewidth Enhancement Factor
dt
P

• Time-dependent frequency change, Dn(t), of laser

- a lw é d (ln P (t )) ù
Dn (t ) = ( ) • Function of output optical power P(t)
+ kP t
4pG êë dt ú
û • Frequency chirp – a function of alw

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide5
 Reducing the Effects of Chirp
• Bias laser sufficiently above threshold
– So that modulation current does not drive laser below threshold
– But leads to an extinction ratio power penalty at receiver
• Extinction ratio – the ratio between the ‘on’ state and the
‘off’ state
• Damping of the relaxation oscillations that occur at turn-on
– Has been achieved by shaping the electrical drive pulses
– Tedious and device-specific
• Use QW lasers, multielectrode DFB lasers, Bragg wavelength tuned
DFB laser – that proved to have better chirp reduction
– However such devices require complex fabricational processes
• Allows laser to emit continuously and impress data onto the optical
carrier using external modulator
Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide6
 Limitation of Direct Modulation of Lasers
• Bandwidth
• Frequency chirping – causing the instantaneous frequency of the
output light to vary
• Pulse modulation
– Frequency chirping leads to large frequency shift at the leading
and trailing edges of each pulse
• For multi-mode lasers (FP), modulation can cause significant power
fluctuation of side modes
– Mode partition noise

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide7
 External Modulation of Lasers

• We have already seen how direct modulation of semiconductor lasers is

limited
– on-off transitions times (max bit rate)
– frequency chirping
• Why not drive the laser cw, and externally modulate the laser light?
• Desired attributes of external modulators
– large modulation bandwidth - fast response (limited by physical properties)
– large modulation depth - large capacity & higher receiver sensitivity
– high-linearity - analog communications
– small size, low power consumption, low voltage operation - engineering
– polarisation independence - modulate signals with arbitrary polarisation
– low insertion loss - optical power is a valuable commodity

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide8
 External Modulation Techniques

• Variety of processes that can be used to modulate light

– Electro-optic effect - apply electric field to induce constant change in
refractive index
– Electro-absorption effect - apply electric field to change gain properties of
semiconductor material
– Acousto-optic effect - apply sound wave to induce periodic change in
refractive index (i.e. grating)
– Magneto-optic effect - magnetic fields, similar to electro-optic effect
– Thermo-optic affect - temperature sensitivity of optical fibre
– Mechano-optic techniques - physical switching
– All-optical techniques - non-linear optical processes

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide9
 Electro-Optic Effect

• Certain material change their optical properties when subjected to

e- field
• Electro-optic effect is the change in the refractive index resulting
from the application of a DC or low-frequency electric field
• The dependence of the refractive index (n) on applied e-field can be
either
– Varies linearly with applied field – linear electro-optic effect
(Pockel effect)
– Varies to the square of the applied field – quadratic electro-
optic effect (Kerr effect)

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide10
 Refractive Index of Electro-Optic Medium

• Refractive index of an electro-optic medium is a function of applied e-field – n(E)

– Varies slightly with E
– Can be expanded is a Taylor’s series about E = 0

1
n(E ) = n + a1 E + a2 E 2 + ...
2
– If y = -2a1/n3 and x = -a2/n3 then

1 1
n(E ) = n - yn 3 E - xn3 E 2 + ...
2 2
– where y is the Pockels coefficient – dominant term in most material
– and x is the Kerr coefficient – significant only when Pockels effect ~ 0

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide11
 Longitudinal and Transverse Mode

• The electric field can be applied to the electro-optic medium in one of 2 ways
– longitudinal - E field applied along the axis of propagation
– transverse - E field applied normal to axis of propagation

L VB
E= VB
L d E=
d

longitudinal transverse
2pL
f = n(E )k0 L =
2pL
(n + a1 E ) = 2pLn + 2pL a1E f = n(E )k0 L = (n + a1 E ) = 2pLn + 2pL a1E
l l l l l l
2pLn 2pL VB 2pLn VB2pL
= + a1 = + a1
l l L l l d
2pLn 2p 2pLn 2p L
= + aV = + a1VB
l l 1 B l l d

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide12
 Longitudinal and Transverse Modes
• Modulation depth (amount of phase change)
– phase change for given L for longitudinal mode increases linearly with VB ,
and since a1 is small, VB >> 1
– if L >> d, then the phase change in the transverse configuration can be large, even
for small VB
• Modulation bandwidth limited by transit time through modulator and RC time constant
– phase transit delay - time it takes for light to propagate from one end of the
modulator to the other
– EO modulator is a capacitor - electrodes separated by dielectric material
• longitudinal - C proportional to d/L, transverse - C proportional to L/d
• increasing the modulation depth also increases the capacitor charge time,
so modulation depth inversely proportional to switching speed
• Electro-optic modulators are polarisation-sensitive
– Align input light to the a polarisation axis with the highest electro-optic effect

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide13
 Phase Modulator

• When light traverses a Pockels medium of length L to which an E is applied, it

undergoes a phase shift, f
öæ 2p ö
æ1
f = n(E )k 0 L = f0 - ç yn 3 E ÷ç ÷L L
è2
øè l ø d
yn 3 EL yn 3VL
= f0 - p = f0 - p
l ld

V
f = f0 - p
• Comparing Vp we see that

• Half-wave Voltage
d l
Vp = • Applied voltage at which the phase shift changes by p
L yn 3 • Important modulator characteristic
• Depends on material properties, wavelength and d/L

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide14
 Intensity Modulator – Wave Interference
• Phase delay alone does not affect the intensity of the light beam
• Intensity modulator – use a phase modulator in an interferometer
Ei E1 Eo= E1+ E2

V E1 = Ei e - jf1
E2 E2 = Ei e - jf2

• Beamsplitters divide the optical power equally

• Constructive and destructive superposition of E fields occurs at output
• Output e-field, Eo
Eo = Ei e - jf1 + Ei e - jf2

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide15
 Intensity Modulator –Wave Interference 2

• Eo can be simplified to

Eo = Ei e - jf1 + Ei e - jf2 = Ei (cos f1 - j sin f1 + cos f 2 - j sin f2 )

= Ei (cos f1 + cos f 2 - j (sin f1 + sin f 2 ))
æ æ f + f2 ö æ f - f2 ö æ æ f1 + f 2 ö æ f1 - f 2 ö ö ö÷
= Ei çç 2 cosç 1 ÷ cosç 1 ÷- j çç 2 sin ç ÷ cosç ÷ ÷÷ ÷
è è 2 ø è 2 ø è è 2 ø è 2 øøø
æ f - f 2 öæ æ f + f2 ö æ f + f2 ö ö
= 2 Ei cosç 1 ÷çç cosç 1 ÷ - j sin ç 1 ÷ ÷÷
è 2 øè è 2 ø è 2 øø
æ f1 +f 2 ö
æ f1 - f 2 ö - j çè ÷
Phase difference between paths determines
Eo = 2 Ei cosç ÷e
2 ø

è 2 ø output light intensity

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide16
 Intensity Modulator – Wave Interference 3
• Output optical power, Po can be obtained via

æ f - f2 ö æ Df ö where Df = f - f
Po = E Eo = 4 E cos ç 1
*
÷
2 2 Po = Pi cos 2 ç ÷ 1 2
o
è 2 ø
i
è 2 ø

Pi A P0
• Output optical power from the intensity
modulator varies with the relative phase
Output difference (Df)
Power
B • Operation:
– The modulator can be operated in the
linear region – adjusting Df = p/2
Df (quadrature point, B)
p C – Or switches light on and off as V switches
between 0 and Vp (between point A and C)

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide17
 Electro-Absorption Modulator
• Electro-absorption (EA) modulator has similar structure as a semiconductor laser –
both can be integrated on the same chip
• EA modulator
– Reverse biased – absorbs light at laser wavelength
– Unbiased – transparent – allows light to pass through
• Laser and modulator are electrically isolated from each other
• Steady current drives the laser, input signal drives the modulator
EA
Laser modulator
Forward bias Reverse bias/ unbias

DFB gratings

active layer substrate Modulated output

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide18
Photo-detectors
 Principle of Light Detection
• Photogeneration of en electron-hole pair
• Condition: hn ³ Ec-Ev
Ec
Ec

hn
Ev Ev
Incident photon excite electron Photogenerated eletron-hole
to make the transition pairs separated

• If we can stop the recombination by making the carriers

move in the opposite direction, we will have current
proportional to incoming photon count
Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide20
 Photodiode with p-n Junction

• Provide a reverse bias on the p-n junction so that

electrons and and holes can be quickly separated
p n by the potential difference caused by the reverse
bias

Depletion
Region • Electrons move to n side and holes move to p side

E cp
• Absorption of photons occurs mainly within
E Fp E(Vo+V) the depletion region
E vp E cn
E Fn • The transport process of the electron-hole
pair generates a current - photocurrent
E vn
Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide21
 Performance Parameters
Number of electron-hole pairs generated and
Quantum Efficiency (h) = collected
Number of incident photons
Ip /q
h=
Po / hn
Ip – Photo current produced by the photodiode
P0- Incident Optical Power

Photocurrent (Ip)
Responsivity (R, A/W) =
Incident Optical Power (P0)

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide22
 Responsivity vs Wavelength 1
P0
rp =
hn
re = hrp = h
P0 hq hql
Ip
R= = =
hn P0 hn hc
P
Ip = re q = h 0 q
hn

• The responsivity, R, is a linear function of quantum

efficiency for a given wavelength. For ideal
photodiode, R is also directly proportional to
wavelength
• Quantum efficiency, h is a function of wavelength, h (l )ql
therefore R is R ( l ) =
hc
Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide23
 Responsivity vs Wavelength 2

• It can be seen that, there is an Responsivity (A/W)

upper bound beyond which 1
R decreases rapidly, lc – 0.9
cutoff wavelength 0.8 Ideal Photodiode
0.7 QE = 100% h( = 1)
hc
lc = 0.6
Eg 0.5
lc
0.4
• For too short wavelength, 0.3 Si Photodiode
photons are absorbed at the 0.2
starting end and then the 0.1
generated electrons and holes 0
0 200 400 600 800 1000 1200
recombined Wavelength (nm)

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide24
 Responsivity vs Wavelength 2
• It can be seen that, there is Responsivity (A/W)
an 1
upper bound beyond which 0.9
R decreases rapidly, lc – 0.8 Ideal Photodiode
cutoff wavelength 0.7 QE = 100% h( = 1)
hc
lc = 0.6
Eg 0.5
lc
0.4
• For too short wavelength, 0.3 Si Photodiode
photons are absorbed at the 0.2
starting end and then the 0.1
generated electrons and holes 0
0 200 400 600 800 1000 1200
recombined
Wavelength (nm)

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide25
 p-i-n Diodes
• Efficiency can be improved by extending the depletion region
• Depletion region can be extended by introducing an intrinsic
(or lightly doped n-type) material between p and n layers
– p-i-n photodiode
Electric Field
pp
Absorption and
i
i depletion
regions
n+

Guest Lecturer from the Dept. of Electrical and ©Thas A Nirmalathas 2016 Optical Communications Systems:
Electronic Engineering, The University of Melbourne LectureSet4_slide26
 PIN - Speed Response
• Time for a carrier to drift across the depletion region
– most fundamental limitation
– increase the electric field to increase the speed of
carriers
– reduce the intrinsic region thickness
– increases the capacitance
• Time for carriers generated outside the depletion region to
diffuse out
– region where absorption can take place but outside
the depletion regions
– diffusion is much slower process (no electrical field)
• Time constant incurred by capacitance of
photodiode
• Small total time constant gives high bandwidth
©Thas A Nirmalathas 2016 Optical Communications Systems
Guest Lecturer from the Dept. of Electrical and Electronic :
Engineering, The University of Melbourne LectureSet3_slide27
 PIN – Pulse Responses

• When the depletion region width w is much larger than the inverse of
absorption coefficiency as (most light will be absorbed), small rise and fall
times of PIN, output follows input well (b)
• When PIN capacitance is larger, then response time is limited by the RC
constant (c)
• Depletion region is too narrow, short drift time, while carriers created outside
the depletion region will have to diffuse back to the depletion region –
response is a slow decaying tail (d)
Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide28
 Avalanche Photodetectors
• Apply a large reverse bias on a photodetector, creates a very high electric field
– Photo-generated electrons accelerated and undergoes ionising collision with
lattice atoms creating a secondary electron-hole pair - impact ionisation
– the reaction continues to build up more and more electron-hole pairs – this
phenomenon is called the avalanche effect

– can realise a gain in the photogenerated current

– if the reverse bias is above a critical voltage (slightly less than break
down voltage), then the multiplication can be continuously realised
Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide29
 Avalanche Photodiodes (APD)

• Response is limited by:

Electric
– transit time of carriers
p+ Field across absorption region
– time for carriers to
perform avalanche
i Absorption
multiplication
region
p+ – RC time constant
n Gain resulting from junction
region capacitance and load
impedance

• APD is realised by separating the absorption and gain regions

– Photon fully absorbed in intrinsic layer (called i region or p region)
– Avalanche effect takes place in the pn+ junction where high e-field exists

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide30
 Multiplication Factor or Avalanche Gain

• A measure of the internal gain provided by the avalanche process

Total output current
I output
M = at the output of APD
Ip
Multiplication Factor Primary photocurrent
(before multiplication)

• Apart from this factor of gain, responsivity and efficiency relationships are
analogousI as for PIN
MIdiodes Mhq Mhql
output p
RAPD = = = = = RM
P0 P0 hn hc • R is the unity gain
responsivity

Guest Lecturer from the Dept. of Electrical and ©Thas A Nirmalathas 2016 Optical Communications Systems:
Electronic Engineering, The University of Melbourne LectureSet4_slide31
 APD vs PIN

• APD • PIN
– internal gain provides an – no internal gain
improvement of 5-10 dB in
sensitivity – simpler structures
– more complex structures – low noise
– random nature of M leading – low bias voltage sufficient
to excess noise – larger bandwidth
– require high bias voltages
– M is a function of
temperature
– good for low speed operation

Guest Lecturer from the Dept. of Electrical and ©Thas A Nirmalathas 2016 Optical Communications Systems:
Electronic Engineering, The University of Melbourne LectureSet4_slide32
 Equivalent Circuits
Ideal Diode and
Capacitance
• p-i-n photodiode
– Reverse biased junction
Light
exhibits a strong
RL capacitance (Cj)
Cj
• APD
– In addition to the
capacitance of the junction,
additional time delays will
effectively lead to much
larger capacitance

Guest Lecturer from the Dept. of Electrical and ©Thas A Nirmalathas 2016 Optical Communications Systems:
Electronic Engineering, The University of LectureSet4_slide33
 Noise Processes in Photodiodes
• PIN diodes
– dark current noise
– quantum noise
• APD
– dark current noise
– quantum noise
– avalanche noise
• whenever a Photodiode is connected to a load resistance, thermal
noise should also be taken into account

• whenever there is a background radiation (from sun), that

contribution also needs to be taken into account

Guest Lecturer from the Dept. of Electrical and ©Thas A Nirmalathas 2016 Optical Communications Systems:
Electronic Engineering, The University of LectureSet4_slide34
 Dark Current Noise
• Electron-hole pairs generated in the absence of input light
– thermal agitation
– resulting current is like noise in the measurement.
• Measures inherent electrical noise within the detector
– Sets a floor on the minimum detectable signal – for the optical signal
to be detected it must produce more current than the dark current
• The process of thermally assisted electron-hole pair generation leading to
dark current is a statistical event
• ie. There is a fluctuation around the mean value of ID
• Mean square dark current noise is given by:

iD2 = iD2 = 2 × q × B × I D
B is determined by photodetector bandwidth and followed low pass filter
bandwidth
Guest Lecturer from the Dept. of Electrical and ©Thas A Nirmalathas 2016 Optical Communications Systems:
Electronic Engineering, The University of Melbourne LectureSet4_slide35
 Quantum Noise/ Shot Noise
• The process of absorption leading to photo current is also
a statistical event
• ie. there is a fluctuation around the mean value of IP
• The mean square quantum noise current

is2 = is2 = 2 × q × B × I p
Quantum
noise Average
current Photocurrent

Total 2
itotal _ PIN = iD
2
+ is2 = 2 × q × B × (I D + I p )
Noise
Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide36
 Avalanche Noise ( for APDs only)

• The process of avalanche multiplication is again a random

process and only applies to APD
• Basically the total of dark current and shot noise is
multiplied by the excess noise factor
• Mean square noise current from APD is given by:

2+ x
< itotal
2
_ APD >= 2 × q × B × ( I D + I P ) × M
Total noise
current Excess Noise
Factor

Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide37
 Thermal Noise

• We also need to take into account the thermal noise of the

receiver circuitry.
• The photodetector load resistor contributes a mean-square
thermal noise current given by

4kT
<i 2
thermal >= B
Thermal
RLOAD
Noise
Current Load
Resistance

Guest Lecturer from the Dept. of Electrical and ©Thas A Nirmalathas 2016 Optical Communications Systems:
Electronic Engineering, The University of LectureSet4_slide38
 Signal-to-Noise-Ratio (SNR)

• SNR of the photodetector is defined as

i 2p
SNR = Mean square
photocurrent
i 2
shot + i 2
dark + i 2
thermal

I 2p M 2
SNR =
( )
2qB I D + I p M 2+ x + 4kTB Rload

• PIN (M=1) photoreceiver is dominated by thermal noise

• APD receiver can be dominated by shot noise
Guest Lecturer from the Dept. of Electrical and ©Thas A Nirmalathas 2016 Optical Communications Systems:
Electronic Engineering, The University of LectureSet4_slide39
 Trans-impedance Amplifier
ii io
+ +
Rmocii
vi
Ri Ro vo
- -

• Trans-impedance amplifiers (current into voltages) are often

required to convert weak photo-currents from the
photodiodes into a useful signal to drive other electronic
detection circuits.
• Some detectors will have these integrated into their
packages.
Guest Lecturer from the Dept. of Electrical and Electronic ©Thas A Nirmalathas 2016 Optical Communications Systems:
Engineering, The University of Melbourne LectureSet4_slide40
 Trans-impedance amplifier
Rf

Cf
R v1
-
Cj
iin
+ vo +
-

Guest Lecturer from the Dept. of Electrical and ©Thas A Nirmalathas 2016 Optical Communications Systems:
Electronic Engineering, The University of Melbourne LectureSet4_slide41